Using Cursor (and other AI code gen tools) for data science
summary
Claims that Cursor and similar AI IDEs produce production‑quality data‑science code while also failing on important tasks. Demos use Cursor, Claude, Bokeh and Streamlit, plus sklearn examples: recreated a 3‑week Streamlit UI in 30 minutes, logistic regression model in ten minutes, but frame and aggregation work on Pandas is brittle — dataframe manipulation fails. Notes cost ($20/month for day‑to‑day work), non‑deterministic quality by time of day, and dangerous git automation that can wipe repositories.
AI-written summary · about
chapters (8)
- 0:00AI code generation for data science
- 3:11Cursor overview and workflow
- 11:17Cursor for visualization code
- 16:36Streamlit UIs with AI
- 21:49Cursor for statistical problems
- 26:22ML model generation in minutes
- 30:15Code generation strengths and limits
- 32:45Q&A
related talks
01h 48mTesting and Securing AI generated code with Cursor
34mUnsession: "AI coding tools"
37mSoftware Engineering + AI = ? | The Future of Development
46mDaily Workflows with Cursor IDE
47mAI-Powered Software Development From the Trenches
36mGenerative AI + Python: Unlocking Efficiency, Personalization, and Insight
16mBuilding Cursor Composer
28mThe Coding Portion of Data Science